Black Friday33% off all plans See plans Enter it at checkout

Prompt Enhancer

The Prompt Enhancer reads your reference pictures with a chat model and writes a detailed prompt from them: the subject, the camera, the light, the setting and the mood, so the generator has more to work with. You can add ready-made cues for light, lenses, cameras and more, edit the result, and copy it into the prompt box.

When to use it

  • The reference is rich (a film still, a moodboard pin) and you want the generator to match its style.
  • Your prompt is short (“woman walking through forest”) and you want concrete detail filled in.
  • You are iterating on a generation and want the prompt to follow the references.

Open it

In a generation panel (Generate Image, Generate Video or Generate Voice), click the sparkle button (Prompt Enhancer) on the bottom bar. The Prompt Enhancer panel opens beside the generation panel; click it again, or the panel’s ✕, to close it.

Build a prompt from your references

  1. Media Inputs — the pictures already in the generation panel’s image slots are copied in when the enhancer opens. Add more from the board, the media library or your computer, or remove the ones you don’t want analysed.
  2. Enhancer Model — pick the chat model that reads the pictures. The list has the chat models of the keys you have set (OpenAI, Anthropic, Google, WaveSpeed or Pika). With no chat key it reads “No API keys configured”. If you add a clip, only models that can read video stay in the list; with none it reads “No model supports current media mix”.
  3. Optional guidance — the field under the model takes notes such as the product style, the brand tone or the target mood.
  4. Click Analyze content and build prompt. The model looks at your media and your guidance, and writes the prompt into Enhancer Prompt (Editable).
  5. Edit the prompt as you like, then click Copy to prompt area. It replaces the text in the generation panel’s prompt box and is also copied to the clipboard.

Prompt Suggestions

Under the prompt, Prompt Suggestions offers ready-made cues in tabs: Light, Environment, Lenses, Cameras, Style, Objects and Colors. Click a suggestion to add it to the end of Enhancer Prompt (Editable) as its own line, tagged with its tab (for example “[Light] golden hour rim light, soft shadows, warm bounce fill”). Click it again to take it out.

You can use suggestions on their own, without analysing any media, and then copy the result to the prompt area.

Dictate instead of typing

The guidance field and the editable prompt each have a microphone. See Voice Input.

Cost

Analysing your media is one request to the chat model you picked, billed by its provider like any other call — usually a fraction of a cent. It isn’t part of the price shown next to Generate, which is the generation’s own price.

Where to go next

Was this helpful?

0

Updated

Comments

Leave a Reply